Effective Human Motor Imagery Recognition via Segment Pool Based on One-Dimensional Convolutional Neural Network with Bidirectional Recurrent Attention Unit Network

Brain–computer interface (BCI) technology enables humans to interact with computers by collecting and decoding electroencephalogram (EEG) from the brain. For practical BCIs based on EEG, accurate recognition is crucial. However, existing methods often struggle achieve a balance between accuracy complexity. To overcome these challenges, we propose 1D convolutional neural networks bidirectional recurrent attention unit network (1DCNN-BiRAU) random segment recombination strategy (segment pool, SegPool). It has three main contributions. First, SegPool proposed increase training data diversity reduce impact of single splicing method model performance across different tasks. Second, it employs multiple CNNs, including local global models, extract channel information simplicity efficiency. Third, BiRAU introduced learn temporal identify key features in time-series data, using forward–backward an gate RAU. The experiments show that our effective robust, achieving 99.47% 91.21% binary classification at individual group levels, 90.90% 92.18% four-category classification. Our demonstrates promising results for recognizing human motor imagery potential be applied scenarios such as brain–computer interfaces neurological disorder diagnosis.